You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
部署时一共遇到了三个问题, 问题 1: ImportError: cannot import name 'cached_download' from 'huggingface_hub' 原因:requirements.txt 中要求安装 diffusers==0.27.2,但 diffusers 的依赖项 huggingface-hub 未指定版本。
未指定版本,则 pip 默认会安装符合要求的最新版本(我此时是0.26.x),但是 huggingface-hub 从 v0.26.0 版本开始就移除了 cached_download() 方法。
( 在 huggingface-hub v0.26.0 的 Release Notes (https://github.com/huggingface/huggingface_hub/releases/tag/v0.26.0) 里可以查到:
Breaking changes
A few breaking changes have been introduced:
cached_download(), url_to_filename(), filename_to_url() methods are now completely removed.From now on, you will have to use hf_hub_download() to benefit from the new cache layout. ) 解决方案: 降低 huggingface-hub 的版本,例如降级到 v0.25.0: pip install huggingface-hub==0.25.0
部署时一共遇到了三个问题,
问题 1: ImportError: cannot import name 'cached_download' from 'huggingface_hub'
原因:
requirements.txt
中要求安装diffusers==0.27.2
,但diffusers
的依赖项huggingface-hub
未指定版本。未指定版本,则 pip 默认会安装符合要求的最新版本(我此时是0.26.x),但是
huggingface-hub
从 v0.26.0 版本开始就移除了cached_download()
方法。( 在 huggingface-hub v0.26.0 的 Release Notes (https://github.com/huggingface/huggingface_hub/releases/tag/v0.26.0) 里可以查到:
Breaking changes
A few breaking changes have been introduced:
cached_download(), url_to_filename(), filename_to_url() methods are now completely removed.From now on, you will have to use hf_hub_download() to benefit from the new cache layout. )
解决方案: 降低
huggingface-hub
的版本,例如降级到 v0.25.0:pip install huggingface-hub==0.25.0
问题 2: ImportError: cannot import name 'EncoderDecoderCache' from 'transformers'
原因:
requirements.txt
中指定了transformers
的版本,但未限制peft
的版本。这导致 pip 自动安装了较新版本的 peft(我此时是 v0.14.0),而旧版本的
transformers
与新版本的peft
可能不兼容。解决方案:降低
peft
的版本,降到0.13.0:pip install peft==0.13.0
问题 3: 运行
gradio_demo.py
,可以正常打开可视化页面,但是推理图片时报 error具体原因不清楚,但是改了一下包的版本后解决了.....
解决方案:把
gradio
的版本升到3.48.0pip install gradio==3.48.0
The text was updated successfully, but these errors were encountered: